|
|
--- |
|
|
license: apache-2.0 |
|
|
base_model: cognitivecomputations/dolphin-2.6-mistral-7b |
|
|
tags: |
|
|
- cybersecurity |
|
|
- security-research |
|
|
- dolphin |
|
|
- mistral |
|
|
- unsloth |
|
|
--- |
|
|
|
|
|
# Dolphin Cybersecurity Research Model |
|
|
|
|
|
Fine-tuned Dolphin-Mistral-7B model for cybersecurity research and education. |
|
|
|
|
|
## Model Details |
|
|
|
|
|
- **Base Model**: Dolphin 2.6 Mistral 7B |
|
|
- **Training Data**: General Knowledge dataset (37,635 examples) |
|
|
- **Training Method**: LoRA fine-tuning with Unsloth |
|
|
- **Use Case**: Cybersecurity education, penetration testing methodology, security research |
|
|
|
|
|
## Usage |
|
|
```python |
|
|
from transformers import AutoModelForCausalLM, AutoTokenizer |
|
|
|
|
|
model = AutoModelForCausalLM.from_pretrained("YOUR_USERNAME/dolphin-cybersec") |
|
|
tokenizer = AutoTokenizer.from_pretrained("YOUR_USERNAME/dolphin-cybersec") |
|
|
|
|
|
prompt = """Below is an instruction that describes a task, paired with an input that provides further context. Write a response that appropriately completes the request. |
|
|
|
|
|
### Instruction: |
|
|
What is SQL injection? |
|
|
|
|
|
### Input: |
|
|
|
|
|
|
|
|
### Response: |
|
|
""" |
|
|
|
|
|
inputs = tokenizer(prompt, return_tensors="pt") |
|
|
outputs = model.generate(**inputs, max_new_tokens=512) |
|
|
print(tokenizer.decode(outputs[0])) |
|
|
``` |
|
|
|
|
|
## Training Details |
|
|
|
|
|
- **LoRA Rank**: 16 |
|
|
- **Learning Rate**: 2e-4 |
|
|
- **Training Steps**: 1000 |
|
|
- **Batch Size**: 2 (gradient accumulation: 4) |
|
|
|
|
|
## Intended Use |
|
|
|
|
|
Educational and research purposes only. For learning about cybersecurity concepts and methodologies. |
|
|
|
